We determined whether the apneic threshold after active hyperventilati
on was different in rapid-eye-movement (REM) vs. non-REM (NREM) sleep.
Sleeping dogs were repeatedly exposed to 35-45 s of hypoxia of varyin
g severity (end-tidal PO2 40-60 Torr) that was abruptly terminated wit
h 100% O2. Changes in breathing pattern after brief hypoxia were compa
red with those after a normoxia-to-hyperoxia transition, i.e., control
conditions. In NREM sleep, hypoxic hyperventilation was consistently
followed by central apnea, the duration of which was linearly related
to the corresponding hypocapnia and/or increase in tidal volume (VT) d
uring hypoxia. After hypoxia, expiratory duration averaged 3.5 X contr
ol value at -5-Torr change in end-tidal PCO2 and twofold increase in V
T; mean expiratory duration was 5 X control value at -10-Torr change i
n end-tidal PCO2 and fourfold increase in VT. In REM sleep, central ap
nea of varying duration did occur on occasion after brief hypoxic hype
rventilation, but there was no systematic relationship with magnitude
of hypocapnia or increase in VT. Breathing pattern during or after hyp
oxia in REM was not related to temporal changes in either eye movement
density or electroencephalogram frequency. Thus, in contrast to NREM
sleep, in REM sleep (''phasic'' or ''tonic'') a posthyperventilation a
pneic threshold was not present. We attribute this effect of REM to 1)
a reduced VT response to hypoxia that would minimize inhibitory ''mem
ory'' effect from lung stretch and 2) attenuated inhibitory response t
o any given magnitude of hypocapnia or increased VT. Active hyperventi
lation-induced apneic threshold may be ''masked'' by actions of nonche
moreceptor and nonmechanoreceptor inputs affecting respiratory motor o
utput in REM sleep. These data are consistent with the relative absenc
e of central apnea and periodic breathing in humans in REM sleep.